U.S. patent application number 16/511162 was filed with the patent office on 2019-11-07 for structure condition sensor and remediation system.
The applicant listed for this patent is Hartford Fire Insurance Company. Invention is credited to Casey Ellen Campbell, Joseph R. Carvalko, JR., Kelly L. Frey, Jonathan Helitzer, G. Stewart Murchie.
Application Number | 20190340696 16/511162 |
Document ID | / |
Family ID | 34226344 |
Filed Date | 2019-11-07 |
![](/patent/app/20190340696/US20190340696A1-20191107-D00000.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00001.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00002.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00003.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00004.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00005.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00006.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00007.png)
![](/patent/app/20190340696/US20190340696A1-20191107-D00008.png)
United States Patent
Application |
20190340696 |
Kind Code |
A1 |
Helitzer; Jonathan ; et
al. |
November 7, 2019 |
STRUCTURE CONDITION SENSOR AND REMEDIATION SYSTEM
Abstract
A computer system is configured to, responsive to receipt of a
data alert from a sensor technology incorporated in a structure,
perform real time polling of remedial technology in the structure,
receive remedial action monitoring data output by the sensor
technology and indicative of whether a remedial action has been
taken, and input the monitoring data alert and the remedial action
monitoring data into a trained computerized neural network. The
neural network is configured to determine and operational status of
the sensor technology and whether the remedial action has been
taken. The computer system outputs a notification identifying a
current risk and a determination of whether the remedial action has
deployed.
Inventors: |
Helitzer; Jonathan;
(Simsbury, CT) ; Murchie; G. Stewart; (West
Hartford, CT) ; Frey; Kelly L.; (Nashville, TN)
; Campbell; Casey Ellen; (Rocky Hill, CT) ;
Carvalko, JR.; Joseph R.; (Milford, CT) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Hartford Fire Insurance Company |
Hartford |
CT |
US |
|
|
Family ID: |
34226344 |
Appl. No.: |
16/511162 |
Filed: |
July 15, 2019 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14138819 |
Dec 23, 2013 |
10354328 |
|
|
16511162 |
|
|
|
|
13617326 |
Sep 14, 2012 |
8676612 |
|
|
14138819 |
|
|
|
|
12709117 |
Feb 19, 2010 |
8271303 |
|
|
13617326 |
|
|
|
|
10656479 |
Sep 4, 2003 |
7711584 |
|
|
12709117 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06Q 40/08 20130101 |
International
Class: |
G06Q 40/08 20060101
G06Q040/08 |
Claims
1. A system for processing data comprising: one or more storage
devices storing a database identifying a plurality of sensor
technologies that reduce risk of loss to a structure; one or more
computer processors in communication with the one or more storage
devices, configured to: receive a monitoring data alert
electronically output by a sensor technology incorporated in the
structure, the monitoring data alert indicative of a current risk
relating to the structure; responsive to receipt of the monitoring
data alert, perform real time polling of remedial technology in the
structure for taking remedial action to mitigate the current risk;
receive remedial action monitoring data electronically output by
the sensor technology responsive to the real time polling of the
remedial technology for taking remedial action, the remedial action
monitoring data indicative of whether a remedial action has been
taken to mitigate the current risk; input the monitoring data alert
and the remedial action monitoring data into a neural network
trained on data structures indicative of an unmitigated factor
pertaining to the structure and a mitigated factor based on
incorporation of the sensor technology in the structure; determine,
by the neural network, executed by the one or more computer
processors, an operational status of the sensor technology and data
indicative of whether the remedial action has been taken to
mitigate the current risk; and output, responsive to the neural
network determination of whether output indicating whether the
remedial action has been taken to mitigate the current risk, a
notification identifying the current risk and a determination of
whether the remedial action has deployed.
2. The system of claim 1, wherein the one or more computer
processors are configured to receive the monitoring data alert
responsive to real time polling, by the one or more computer
processors, of the sensor technology incorporated in the
structure.
3. The system of claim 2, wherein the neural network comprises one
of: (a) a multilayer neural net including a neural net input layer
having a plurality of computational processing units having a
one-to-one correspondence to technology mitigation data; (b) a
technology classifier having a set of technology mitigation
vectors, each vector specific to one building configuration, and
inputs having a one-to-one correspondence with classification of
mitigation technology values; and (c) a decision tree classifier
developed during a construction phase requiring that a set of
building structures and corresponding combination of technologies
by recursively partitioned into two or more subpartitions.
4. The system of claim 1, further comprising a data storage device
storing data indicative of terms of a risk mitigation policy issued
to a covered entity and covering the structure; wherein the one or
more computer processors are further configured to determine, based
on the operational status of the sensor technology and the data
indicative of the terms of the risk mitigation policy in the
database, whether an alteration in the data indicative of at least
one of the terms of the risk mitigation policy, after issue of the
risk mitigation policy, is warranted and, responsive to a
determination that the alteration in the data indicative of at
least one of the terms of the risk mitigation policy is warranted,
alter data indicative of the at least one of the terms.
5. The system of claim 1, wherein the monitoring data alert
comprises structure condition data and alert data concerning the
current risk.
6. The system of claim 1, wherein the operational status of the
sensor technology comprises one or more of an operational status of
a sensor for sensing the current risk to the structure, an
operational status of an electronic data output for reporting the
current risk to the structure, and an operational status of an
automatic remediation system for deploying the remedial action to
mitigate the current risk to the structure.
7. The system of claim 1, wherein the sensor technology
incorporated in the structure comprises one of smoke detectors,
fire detectors, intrusion systems, radiation detectors, chemical
detectors, biological hazard detectors, water level detectors,
water leakage detectors, vibration detectors, and meteorological
condition detectors.
8. The system of claim 1, wherein the remedial action monitoring
data comprises at least one of: data from a chemical remediation
system for mitigating a mold risk to the structure; data from a
water pumping remediation system for mitigating a water risk to the
structure; and data relating to a type of sprinkler system for
mitigating a fire risk to the structure.
9. The system of claim 1, wherein the structure comprises one of a
commercial building, a residential building, a vehicle, a marine
craft, and an aircraft.
10. A computer-implemented method comprising: maintaining in a
database stored in one or more storage devices, data identifying a
plurality of sensor technologies that reduce risk of loss to a
structure; receiving, by one or more computer processors, a
monitoring data alert electronically output by a sensor technology
incorporated in the structure, the monitoring data alert indicative
of a current risk to the structure; polling, in real time by the
one or more computer processors, remedial technology in the
structure for taking remedial action to mitigate the current risk;
responsive to the real time polling of the remedial technology for
taking remedial action, receiving, by the one or more computer
processors, remedial action monitoring data electronically output
by the sensor technology indicative of whether a remedial action
has been taken to mitigate the current risk; inputting the
monitoring data alert and the remedial action monitoring data into
a neural net trained on data structures indicative of an
unmitigated factor pertained to the structure and a mitigated
factor based on incorporation of the sensor technology in the
structure; determining, by the neural network, executed by the one
or more computer processors, an operational status of the sensor
technology and data indicative of whether the remedial action has
been taken to mitigate the current risk; and outputting, responsive
to the neural network determination of whether output indicating
whether the remedial action has been taken to mitigate the current
risk, a notification identifying the current risk and a
determination of whether the remedial action has deployed.
11. The method of claim 10, wherein receiving the monitoring data
alert is responsive to real time polling of the sensor technology
incorporated in the structure.
12. The method of claim 10, wherein the neural network comprises
one of (a) a multilayer neural net including a neural net input
layer having a plurality of computational processing units having a
one-to-one correspondence to technology mitigation data; (b) a
technology classifier having a set of technology mitigation
vectors, each vector specific to one building configuration, and
inputs having a one-to-one correspondence with classification of
mitigation technology values, or (c) a decision tree classifier
developed during a construction phase requiring that a set of
building structures and corresponding combination of technologies
by recursively partitioned into two or more subpartitions.
13. The method of claim 10, further comprising: storing, in a data
storage device, data indicative of terms of a risk mitigation
policy issued to a covered entity and covering the structure; and
determining, based on the operational status of the sensor
technology, the data indicative of whether the remedial action has
been taken to mitigate the current risk, and the data indicative of
the terms of the risk mitigation policy, whether an alteration in
the data indicative of one of the terms of the risk mitigation
policy is warranted and, responsive to a determination that the
alteration in the data indicative of at least one of the terms of
the risk mitigation policy is warranted, altering data indicative
of the at least one of the terms.
14. The method of claim 10, wherein the monitoring data alert
comprises structure condition data and alert data concerning the
current risk.
15. The method of claim 10, wherein the operational status of the
sensor technology comprises one or more of an operational status of
a sensor for sensing the current risk to the structure, an
operational status of an electronic data output for reporting the
current risk to the structure, and an operational status of an
automatic remediation system for deploying the remedial action to
mitigate the current risk.
16. The method of claim 10, wherein the sensor technology
incorporated in the structure comprises one of smoke detectors,
fire detectors, intrusion systems, radiation detectors, chemical
detectors, biological hazard detectors, water level detectors,
water leakage detectors, vibration detectors, and meteorological
condition detectors.
17. The method of claim 10, wherein the remedial action monitoring
data comprises one of: data from a chemical remediation system for
mitigating a mold risk to the structure; data from a water pumping
remediation system for mitigating a water risk to the structure;
and data relating to a type of sprinkler system for mitigating a
fire risk to the structure.
18. The method of claim 10, wherein the structure is one of a
commercial building, a residential building, a vehicle, a marine
craft, and an aircraft.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This is a continuation application of and claims priority to
copending U.S. patent application Ser. No. 14/138,819, filed Dec.
23, 2013, which is a continuation application of and claims
priority to U.S. patent application Ser. No. 13/617,326, filed Sep.
14, 2012, now U.S. Pat. No. 8,676,612, which is a continuation
application of and claims priority to copending U.S. patent
application Ser. No. 12/709,117, filed Feb. 19, 2010, now U.S. Pat.
No. 8,271,303, which is a continuation application of and claims
priority to U.S. patent application Ser. No. 10/656,479, filed Sep.
4, 2003, now U.S. Pat. No. 7,711,584, the entirety of each of the
foregoing applications being hereby incorporated by reference
herein for all purposes.
BACKGROUND OF THE INVENTION
1. Field of the Invention
[0002] This invention relates to a method and computerized system
for creating, publishing, underwriting, selling and managing
insurance products, the issuance criteria and premium, for which,
is based upon the technology utilized in connection with the
insurable interest.
2. Description of the Prior Art
[0003] Underwriting is the process of establishing insurability and
premium levels that will economically and profitably transfer risk
from a policyholder to an insurance company. In determining
insurability and premium, insurance carriers take into account such
factors as profit goals, competition, legal restrictions and the
costs associated with losses (claims costs), loss adjustment
expenses (claim settlements), operational expenses (commission and
brokerage fees), general administrative expenses, and the cost of
capital.
[0004] More particularly, an insurance carrier typically assesses a
unit of exposure on the basis of a premium, known and predicted
exposure, and loss and expense experience. In this manner carriers
establish the basis of potential loss and the general direction of
trends in insurance claim costs. In setting and subsequently
adjusting which risks to underwrite and the premium a carrier
catalogs by time and place, accidents as well as changes in claim
costs, claim frequencies, loss exposures, expenses and premiums;
the impact of catastrophes on the premium rates; the effect of
salvage and subrogation, coinsurance, coverage limits, deductibles,
coverage limitations or type of risks that may affect the frequency
or severity of claims; changes in the underwriting process, claims
handling, case reserves and marketing practices that affect the
experience; impact of external influences on the future experience,
including the judicial environment, regulatory and legislative
changes, fund availability, and the modifications that reflect the
impact of individual risk rating plans on the overall experience.
However, notably absent from the factors customarily taken into
account and one of the most profound influences in loss experience
is the effect of technology. Therefore, an underwriting process
that considers the continuing technology revolution would be
anticipated to better assess loss ratios for insurable
interests.
[0005] It is widely assumed that using various technologies may
reduce the risks of loss associated with building structures,
generally. Consequently, state, local and national construction
codes affecting such things as structural requirements, electrical
standards, plumbing and paint are constantly being advanced. For
example, home building codes throughout the United States have
placed a minimum standard on requirements for construction to
assure a minimally safe habitation and structural integrity under
typical local conditions for geologic and meteorological
occurrences. Beyond the requirements imposed through legal
regulation, property owners may also employ systems that further
militate against one or another loss or hazard. For well over a
generation, home security systems have been utilized to reduce
losses from dwelling break-ins. Many fire alarm systems
automatically call fire stations minimizing fire damage and
reducing human loss through early detection and central alarm at
the appropriate responder location. In addition to fire
notification, flame-retardants are in wide use to reduce damage
from fire. In some instances one technology replaces another as to
improve a condition that is inherently dangerous, but the
replacement technology retains the fundamental objective of
reducing damage. For example, asbestos has been virtually banned as
a building material in favor of flame retarding products as a means
for reducing fire hazards. As new hazards are discovered, newer
technology will be incorporated to achieve the benefits of a safer
society.
[0006] From a baseline related to minimum code requirements,
underwriters of property and casualty insurance factor into the
risk/loss proposition items that relate to the structure to be
insured (by way of example, the year of construction, type of
construction, area, roof materials, egresses, the property's
physical address, its proximity to fire apparatuses such as fire
hydrants or fire stations, proximity to environmental hazards, such
as superfund dumps, large bodies of water and its current market
and replacement value). Underwriters also take into account items
not directly related to the physical properties of the insurable
interest, but that have been statistically shown to correlate with
risk/loss (by way of example, the insured's credit rating, age of
the property owners, and the insured's prior claim history).
[0007] Contemporary underwriting practice is typically reduced to a
binary choice to issue or not to issue a policy of insurance based
upon the aggregate of statistically relevant underwriting criteria,
rather than producing insurance products tailored to combinations
of risk reduction technology. As such, the benefits of a class of
technology may not be adequately considered during the underwriting
process. Significantly, the range of efficacies associated with
specific technologies within a class of technologies are ignored as
a salient fact.
[0008] A prime example might be an underwriting practice that does
not factor in the functional details of available sensor technology
such as by way of example, the type and corresponding unique
features of the spectrum of smoke detectors, fire detectors,
intrusion systems, radiation, chemical or biological hazard
detectors (such as the detection of disease producing infectious
agents, causing viral infections or the presence of allergens
related to common allergies and forms of sinusitis). Other examples
of sensing potential damaging situations are: water level or
leakage detectors, vibration detectors, and meteorological
conditions.
[0009] Also, the current insurance underwriting practice does not
factor in details on various actively responsive technologies that
are currently available such as by way of example, the type (i.e.
specific functionality) of sprinkler system, the presence of a
chemical release system to, for instance, release fungicides to
kill mold spores or water pumping systems to remove damaging water
or products that communicate medical emergencies. Nor does the
current insurance underwriting practice discriminate between
self-reports by the potential insured of extant technology and
actual, continuous functionality (monitoring) of relevant
technology designed to reduce damage/risks.
[0010] Products as familiar as the common home alarm system or
security lighting, to the less common sensors attached to screens
and windows, would be considered the kinds of products that are
readily available by today's consumer. Overlooked however are the
safety advantages of slip-less floor covering, outside walkways
constructed from materials that insure against the accumulation of
ice and snow and/or have high friction qualities due to the
materials of constructions. However, in the future there will be a
wide variety of products that detect and/or ameliorate all forms of
hazards to property and health that will be available and utilized
depending upon the value, as well as the incentives provided, such
as through less expensive insurance premiums.
[0011] The insurance industry has long recognized the risk
reduction concomitant to the incorporation of certain products in
buildings. Certain products keep loss premium ratios down as well
as providing a benefit to the property owners in terms of reduced
property loss and improved health.
[0012] Thousands of separate and distinct materials and products
are employed in the construction of homes and commercial buildings.
Large numbers of such building products have a significant impact
upon personal safety and the ability of the structure to withstand
catastrophic events. Architects, builders and home owners have
considerable opportunity to chose among diverse products that might
for purposes of discussion be separated into categories such as
building materials, sensor technologies and responder technologies.
An exhaustive list of products from those categories, alone
pertaining to loss prevention and mitigation could reasonably be
expected to run into the millions of combination (e.g. more than
100 different materials times 100 different sensor technologies
times 100 different responder technologies). Various specific
combinations may have corresponding efficacies with regard to the
amelioration of loss. In each instance, the consumer would
anticipate a corresponding premium to reflect the expected loss
ratios attendant to using a particular product or combination of
products (and might be influenced to make more economically sound
judgments in incorporating materials/technologies that reduce
damage/risk, if the benefit of such choices could be clearly
articulated in costs savings from reduced premiums over the life of
the material/technology in question).
[0013] However, the insurance industry generally does not factor
into its actuarial computations or underwriting rules the reduction
in risk with sufficient specificity to affect premiums or expand
coverage that can be underwritten within acceptable loss premium
ratios (either by increasing specificity as to exclusions,
qualifying risk allocation based upon risk reduction technology or
providing extended coverage under excess premium conditions). Nor
does the industry publish or otherwise make available to the
consumer sufficient information on the underwriting process to
allow the consumer to adequately select militating technology that
could result in significant costs savings, both to the consumer and
the insurance underwriter. In as much as classical underwriting
depends to a large degree on statistics surrounding conditions
relevant to loss, the difficulty in utilizing technologic
innovation in the actuarial computations has to do with the small
sample sizes and/or lack of data on the ameliorating effect of a
particular technology.
[0014] As apparent, the salient combination of technologies
utilized in a building may be vast, and searching for specific
combinations and relating them to loss ratio and premiums is a time
consuming process utilizing current information processing systems.
Nonetheless, such systems may feasibly be handled with such
technologies as neural networks as discussed in such patents as
U.S. Pat. Nos. 5,696,907 and 5,893,072.
SUMMARY OF THE INVENTION
[0015] The invention herein disclosed deals with insurance
products, systems and methods that treat the management of risk,
relating to specified, yet unknown, future events. Insurance
underwriters specify a particular product relating to an event or
phenomenon for which there is a range of possible future outcomes.
The related insurance company then offers an insurance contract
relating to the predetermined phenomenon and corresponding range of
outcomes based upon a range of efficacies of suitable technology
within a class of technology. The offered contracts specify a
requirement or an incentive to employ a particular technology to
militate against loss entitlement and specify the maximum policies
insurance limits with and without the militating technology at the
future time a claim is made and a consideration or premium payable
upon binding. Such invention may be expressed in terms of an
underwriting system/process for determining issuance of insurance
and calculation of premium for a single insurable interest or a
system/process for calculation and presentation of multiple
underwriting options/end-states based upon the use of a plurality
of technologies incorporated into the insurable interest to
militate against loss.
[0016] Expert computer systems, which may include decision trees,
neural networks and statistical inference engines, have the ability
to store information, interpret the information and draw inferences
based upon the information such as received from sensors and
descriptors of loss mitigation technology. The general architecture
of an expert system involves a problem dependent set of data
declarations called the knowledge base, and a problem independent
program which is called the inference engine. The data collected
from sensors or other quantifications of loss mitigation technology
such as been described hereinabove can provide the information,
from which data declarations can be constructed and classifications
of the best practice loss mitigation technology for a given
insurable interest may be determined. Pattern recognition systems,
such as utilized in the form of neural networks, provide for a
large theoretical basis for these types of systems. In general see,
Principles of Expert Systems edited by A. Gupta et al., and
published by IEEE Press (1988). By employing an expert system an
individual has the benefit of a system that can provide
qualitative, nontrivial and high quality solutions and answers to
outcomes that depend on complex arrays or quantities of data. See
also, U.S. Pat. No. 5,023,785, R. F. Adrion et. al. entitled
Hematological-Diagnoses Apparatus Employing Expert System
Technology, which describes a system within the class of expert
systems that may be adapted to deal with complex arrays or
quantities of data as may be required in some applications of this
invention. These systems can benefit the analysis of data
collection, analysis, and diagnosis in complex insurance industry
applications, especially when dealing across entire spectrums of
loss mitigation classifications.
[0017] A classifier for improving methods of insurance
underwriting, which includes a processor having one or more inputs
to receive data structures representing a first unmitigated
insurance underwriting risk pertaining to an identified building
structure and a second mollified insurance underwriting risk on the
assumption a certain technology will be employed in the building
structure. The difference between these two data represents the
incorporation of a technology that aides in the reduction of
casualty property losses (either through reduction of claim
incidence or absolute claim costs). Each data structure forms a
logical association that may represent a logical, qualitative,
comparative or quantitative evaluation (collectively hereinafter
referred to as a "difference") which difference may be assigned a
weight referred to as a weighted difference, between the first
field representing an unmitigated risk, and the second field
representing a mollified risk. The plurality of data structures
generate weighted differences to form a plurality of weighted
outputs. At least one of the weighted output signals represents a
class for the unmitigated risk/mollified risk, and the datum
represents an input into a process that sums the weighted
differences to generate a minimized risk for a building structure
under consideration. Essentially, the processor utilizes the
weighted differences to form a logical association that may
represent a logical, qualitative, comparative or quantitative
evaluation (collectively hereinafter referred to as a "sum")
wherein the weighted sum of one or more inputs represents levels of
risk, with and without the technology to reduce the risk. By way of
example, two (2) inputs, and one output might be the simple output
from two risk levels, one with and one without the mitigating
technology. The output is therefore:
W.sub.0*t.sub.0+W.sub.1*t.sub.1+K.sub.b>0 for a risk having two
different technologies incorporated into a hypothetical building
structure and having a constant K resulting from non technology
related losses.
[0018] Alternatively, a classifier for purposes of assigning
combination of technologies existing in a building structure
utilizes a decision tree. In this manner data structures
representing the quantification of risk reduction attendant a given
technology or product can be chained into a plurality of decision
trees. In one aspect of the invention the decision tree provides
for the creation of a decision tree that includes a construction
phase and a pruning phase. The construction phase requires that the
set of building structures and corresponding combination of
technologies be recursively partitioned into two or more
subpartitions until a stopping criterion is met, and a
classification assigned. The decision applies a splitting criterion
to every node of the tree. These splitting criterion are determined
by applying a predetermined rule or function that an underwriter
applies to eventually place the applicant for insurance into a
classification that is then utilized as a factor in establishing
the premium.
[0019] Each node may utilize any number of conventionally available
analytical techniques for producing a splitting criterion. For
example, each branch may be programmed to produce a weighted
average (W.sub.1-N) that may subsequently be applied in successive
nodes and branches to influence the final risk classification.
[0020] In one aspect of the invention, the processor is programmed
to impose a decision criterion based upon whether a set of
technologies exist in a particular type of building structure. The
partitioning eventually leads to specific types of coverages at
specific premium levels, effectively pruning those "risk branches"
that were amenable to insurance coverage under any set of
underwriting parameters from those "risk branches" for which, risk
reduction mechanisms other than insurance are indicated. An
alternate embodiment would incorporate a continuous function for
the limit on W, values of which would selectively result in
specific underwriting choices, exclusions or excess premium
charges.
[0021] The invention herein also discloses a method for managing
the underwriting, quoting and binding of a property and casualty
insurance policy for an insured with regard to the technology used
to militate against certain insurance property losses. More
particularly, the method of underwriting insurance takes into
account technologies that militate against loss the method
comprising the steps of: identifying a technology that improves a
claim risk associated with a property loss against which an insured
purchases insurance; advising the insured to obtain the technology
as a condition of obtaining insurance on the property as well as
advising the insured of the insurance consequences of using
specific technologies, which allows the insured to perform a
cost/benefit analysis; and providing an insurance policy that
accounts for the diminution of risk, or incorporates appropriate
waivers, exclusions or riders, following incorporation of the
technology. Such method may be practiced in conjunction with the
evaluation and underwriting of a specific insurable interest or by
presentation of the plurality of technologies having the greatest
influence on the underwriting criteria and premium calculation in
such a manner as to allow objective analysis of inclusion of a
specific technology (cost) against availability of risk insurance
as a reduced premium over the life of the technology (benefit). The
methods of assessing the best practice loss mitigation technology
also may benefit architects and builders of structures who would
gain design insights into best practices in the construction of
buildings. Such intended effects would have both safety and
economic benefits upon the incorporation of specific
technologies.
[0022] A method for underwriting insurance takes into account
technologies that militate against loss through methodologies
comprising the steps of: maintaining a database identifying a
plurality of technologies that reduce risk of loss suitable for an
associated building structure; maintaining a database of risk
mitigation associated with technologies and the associated building
structures; identifying a building structure comparable to the
associated building structure, that requires insurance; calculating
the risk associated with the building structure that requires
insurance; and accounting for the risk reduction resulting from
incorporation of at least one technology into the building
structure requiring insurance; a means for the creation of a policy
having exclusions appropriate for the structure and the
incorporated risk reduction technologies; and controlling a
printing device to print a complete insurance policy based on the
subject building.
[0023] The invention herein includes an information processing
system for underwriting a potentially insurable risk, comprising:
means providing first and second data bases, a terminal device; a
means for storing information relating to the potentially insurable
risk in the first data base; a means for storing information
relating to the potentially insurable risk mitigated by a
technology in the second data base; a means for evaluating the
information stored in the first data and second data base and for
identifying additional elements of information required for
evaluating the potentially insurable risk, and for requesting entry
of said additional information for subsequent storage in the first
data base; a means for assigning a weight to at least one of the
selected elements of information from the first data base on the
basis of predetermined relationships existing between the elements
of information in the first data base and corresponding elements of
information in the second data base; means for associating selected
elements of information from the first data base with corresponding
elements of information previously stored in the second data base;
means for displaying information corresponding to at least one of
the selected elements of information from the associating selected
elements of information; and means for determining at least one
risk classification for the potentially insurable risk from the
weights assigned to the elements of information in the first data
base.
[0024] The information processing system for underwriting further
comprises: a plurality of terminal means to manage insurance
accounts; a database processor means for storing predetermined
data, said database processor means being interconnected and
responsive to each of said plurality of terminal means; a processor
means for managing the predetermined data, said file processor
means being interconnected and responsive to each of said plurality
of terminal means; an output means for producing documents in at
least one of text, graphics, and electronic transfer mode, said
output means being interconnected and responsive to each of said
plurality of terminal means; and, an input means for receiving
predetermined input data into said information processing system,
said input means being interconnected and responsive to each of
said plurality of terminal means; and, software means for
configuring each of said plurality of terminal means, database
processor means, file processor means, output means, and input
means for, managing said technology insurance underwriting quoting,
policy generation and binding of insurance policies. The system for
managing insurance accordingly may also be adapted to output a
specification of best practices to mitigate losses based upon the
use of selected technologies in the construction utilization within
a subject building.
[0025] The invention herein also discloses a preferred embodiment
for determining the classification of risk and consequent premium
for a structure that utilizes an array of available technologies.
The combinations for such technology driven structures is
characteristically large and the invention herein utilizes a neural
network to determine the optimum risk loss and premium. The neural
network comprises an input layer, an output layer, and at least one
hidden layer, operable to produce outputs from said output layer
when inputs are supplied to the input layer of said neural network,
said artificial neural network having been previously trained in
accordance with training exemplars of loss mitigation due to the
incorporation of certain technologies establishing a particular
knowledge domain therein. Such embodiment may have a resultant
product that informs the production of a specific policy of
insurance or may have a resultant product that quantitatively
presents the economic effects (insurability or reduction in
premium) associated with one or more technologies based upon a
variable (or generic) insurable risk profile (an abstract, rather
than a specific insurable risk).
[0026] The methodology at arriving at the optimum premium comprises
a method for performing risk analysis utilizing a neural network
performing the steps of: collecting data on an efficacy of a first
combination of technologies; applying the data to an input layer of
the network; training the network by optimizing the weighted
connections of the network to an underwriting criteria that
decreases the risk of an insurable interest such that weights are
determined; applying a second combination of technologies to an
input layer of the network classifying the second combination of
technologies into ordinal values and categorical values; such that
the classification represents an optimized premium estimation.
[0027] The invention herein also discloses a preferred embodiment
for determining the classification of risk and consequent premium
for a structure that utilizes an array of available technologies
that are actively monitored for functionality. By way of example,
in the simplest form such active monitoring may be a single or
periodic physical inspection of the technologies as self-reported
by the potential insured, by a representative of the insurance
underwriter, or third party technician with credentials acceptable
to the insurance underwriter to confirm installation and/or proper
function. However, the invention encompasses more sophisticated
real-time polling of the technologies for "current status"
functionality, with the resultant polling report being an active
input into the input layer of the neural net (allowing adjustment
of insurability status and premium calculation based upon the
actual, as opposed to hypothetical, functionality of the
technologies). The invention also encompasses positive prevention
attributes of technologies, as well as simple monitoring, such that
a specific technology may not only provide an alert of a dangerous
condition, but also take immediate remedial actions to militate
such condition with both the alert and remedial action being
subject to real-time polling, such polling being active inputs into
the input layer of the neural net.
BRIEF DESCRIPTION OF THE DRAWINGS
[0028] FIG. 1 shows a block diagram of a generalized computer
system suitable for use in practicing the present invention.
[0029] FIG. 2a and FIG. 2b show a block diagram of the logical
organization of the methodology used in the invention as it relates
to underwriting, quoting, and binding an insurance policy.
[0030] FIG. 3a shows a schematic of the process for classifying
risk input data on the assumption a certain technology will be
employed in the present invention.
[0031] FIG. 3b shows a schematic of the process for classifying
risk input data on the assumption a certain technology will be
employed in the present invention.
[0032] FIG. 4 is a classifier that assists in carrying out the
method of adjusting premium based on mitigation technologies.
[0033] FIG. 5 shows a table of building types and technological
combinations.
[0034] FIG. 6 is a decision tree classifier to carry out the method
of adjusting premium based on mitigation technologies.
DESCRIPTION OF THE INVENTION
[0035] Referring to FIG. 1, the invention disclosed herein is a
system for underwriting, quoting, policy generation and binding an
insurance policy with regard to the technology being employed to
militate against the financial consequences of certain property and
casualty losses. It will be understood that FIG. 1, illustrates an
exemplary embodiment of a system 100 that may be used for
implementing the principles of the present invention. In general,
the system includes a local area network of terminals or
workstations, database file servers, input devices and output
devices configured by software for accumulating, processing,
administering and analyzing insurance in an automated workflow
environment. The system provides, for on-line quoting, rating, and
binding of insurance policies, electronic data transfer and the
evaluation and access to the data relevant to technology pertinent
to reducing costs associated with certain avoidable hazards and
losses. The system also provides for publication of the
quantitative effects of a single or a plurality of technologies
upon the underwriting process based upon-generic inquiries
specifying a range of variables related to technologies
incorporated into the data stored on the database file servers.
[0036] System 100 includes one or more terminals 110a-110n, each
having a corresponding CPU, such as CPU 106 which includes a
display 103 and memory 104. The terminals 110a-110n are used for
underwriting, creating, selling and managing insurance policies,
the issuance and premium of which is based upon the technology
utilized in connection with the insurable interest. In addition, a
database means 150 is interconnected to the terminals 110a-110n for
storing predetermined rate data; output means 160 for producing
documents in at least one of text, graphics, and electronic
transfer mode, said output means being interconnected and
responsive to each of said plurality of terminal means such as
terminal means 110n; and, a corresponding input means 108 for
receiving predetermined input into said CPU 106, and a software
means (unshown) for configuring each of said plurality of terminals
110a-110n.
[0037] Output means 160 represents one or more devices, such as
other facsimile machines, photocopiers, etc., that have access to
rate filings, which may be stored on database 150. Input source 115
also communicates with one or more network connections for
receiving data from a server 140 over network 120, e.g., a global
computer communications network such as the Internet, a wide area
network, a metropolitan area network, a local area network, a
terrestrial broadcast system, a cable network, a satellite network,
a wireless network, or a telephone network, as well as portions or
combinations of these and other types of networks.
[0038] Other servers 140 may be in communication with network 120
or may be in direct communication with terminals 110a-110n. Server
140 and terminals 110a-110n are in communication with database
means 170 to obtain rate information, to store information related
to hazard ratings and for any suitable purpose in underwriting,
creating, selling and managing insurance policies, the issuance and
premium of which, is based upon the technology utilized in
connection with the insurable interest or the publication of the
quantitative effects a single or a plurality of technologies have
upon the underwriting process, and based upon generic inquiries
specifying a range of variables related to the technologies
incorporated into the data stored on the database means 170.
[0039] In addition to on-site databases, such as database means 150
and a remote data base means 170, data may be provided from a data
provider 180.
[0040] In a preferred embodiment, computer readable code executed
by CPU 106 as processing relates to terminals 110n may implement
the coding and decoding employing the principles of the present
invention. Correspondingly, computer readable code executed by
server 140 as processing relates to CPU 145 may implement coding
and decoding employing the principles of the present invention. In
other embodiments, hardware circuitry may be used in place of, or
in combination with, software instructions to implement the
invention. For example, the elements illustrated herein may also be
implemented as discrete hardware elements. As would be appreciated,
terminals 110a-110n and server 140 may be embodied in such means as
a general purpose or special purpose computing system, or may be a
hardware configuration, such as a dedicated logic circuit,
integrated circuit, Programmable Array Logic (PAL), Application
Specific Integrated Circuit (ASIC), that provides known outputs in
response to known inputs. Referring to FIG. 2a and FIG. 2b, a flow
chart illustrates the operation of a computer-implemented software
process 200 for underwriting, quoting, binding, issuing and
managing insurance policies, by an insurance carrier underwriter of
a casualty insurance policy dependent upon the technology existing
within the insurable interest, in accordance with a preferred
embodiment of the present invention. Software process 200 is
preferably implemented on a workstation typical of the terminal
110a-110n such as illustrated in FIG. 1. In the present instance,
the System 100 allows users to access process 200 to perform
underwriting functions; quote policy coverages and premiums for
insurance policies locally and from remote locations.
[0041] Although the following description will refer to a system
for the generation of a commercial property and casualty lines of
insurance for building structures, an equivalent process is
applicable to any insurable interests where the underwriting
criteria and the premium are influenced by the absence or presence
of technology. Such insurable interests are by way of example only,
residential premises, vehicles, marine craft and aircraft. In
referring to FIG. 1 and FIG. 2a, FIG. 2b in a first step 202,
through an input device 108, a user logs into process 200 through
an associated terminal 110n having a display 103, that connects to
a means 150, providing first and second data bases. In the next
step 204, utilizing the input device 108, the user enters quotation
information pertaining to the insured party for whom casualty
insurance dependent upon the technology is to be underwritten. Such
information typically includes, the name, address, telephone number
of the insured party, the date the request for the quotation was
received, a description of the insured's operation and the standard
industrial codes ("SIC"), which are associated with the insured's
business.
[0042] The System 100 and associated software process 200 maintains
several means for storing databases such as means 150 and means
170. As will be apparent to those skilled in the art of programming
the specific storage of databases utilized by the system 100 will
present a design choice for the programmer. The description that
follows may call out a database means attached to a user terminal
such as 110n, but such database means as means 170 attached to
server 140 would also fall within the spirit of the inventions, in
any instance where a database means were required.
[0043] A plurality of SIC records are stored in a database resident
in a database means 150. Each of the SIC records are linked to
underwriting guidelines (unshown) established and filed by the
insurance carrier. These criteria include guidelines related to
minimum premiums, hazard rating, underwriting authority, and
referral criteria.
[0044] The process 200 displays on display 103 a plurality of
candidate risk modifiers, such as may be retrieved at points in the
process as by way of example risk modifier 213, 215 or 217, each
associated with one or more technologies that mitigate the risk of
loss or hazards associated with the insurable business property.
Step 204 records a selected risk modifier code 213 and related
underwriting criteria associated with the business property and
associated policy.
[0045] The process 200 proceeds to step 206, where the user enters
into the process 200 the name of the carrier, and the coverage type
and coverage limits of the insurance policy. Since the underlying
insurance policy may have separate limits for general liability and
specifically named liability coverages, the insurance policy
producer may enter separate primary coverage limits for general
liability and specific liability coverages in this step. In step
206, the insurance policy producer enters the expiration dates of
the proposed insurance contract and a description of the insured
property.
[0046] In step 208, the process 200 retrieves from a first database
resident in database means 150, public bureau rating information.
The present invention maintains in database 150 information
relating to the potentially insurable risk, mitigated by a
technology. In step 210, the rating quoted is compared against a
predetermined minimum technology-rating threshold 210 established
by the carrier issuing the insurance offer quotation. The process
200 takes into account the risk modification step 215, where a risk
modifier code factors into the decision, the effects of mitigation
of risk through the use of technology. If, as a result of this
comparison, the system determines that the rating of the insurance
carrier is below the predetermined threshold, the system proceeds
to step 212, where the insurance underwriter is given the option to
decline to issue a quotation or refer the submission to a managing
authority for further consideration. If the user declines to issue
the quotation in step 212, then the system optionally generates a
declination letter, indicating that no quote will be submitted for
the casualty policy dependent upon technology; otherwise, the
system proceeds to step 214 where the user is typically required to
document reasons for writing coverage that does not meet minimum
underwriting criteria.
[0047] In step 216, the system retrieves and displays underwriting
guidelines associated with the SIC that were previously entered in
step 204. The present invention maintains a database residing on
database means 150, which contains underwriting instructions and
guidelines, including minimum premiums, loss or hazard mitigation
technology and hazard rating instructions, corresponding to each
SIC that a user might enter into the system in step 204.
[0048] The loss or hazard mitigation technology and hazard rating
instructions contain factors that are considered when associating a
risk to a particular SIC. Based on this loss or hazard mitigation
technology and hazard rating information, the user selects one or
more ratings for the quotation in step 218. The selected loss or
hazard mitigation technology and hazard rating(s) are then stored
in the process 200 database means 150 as part of the computer file
associated with the particular quotation. In step 219, the user may
obtain a physical verification of the use of certain risk
mitigation technology for the insurable interest under
consideration.
[0049] In step 220, the user chooses one or more of the coverage
types which are applicable to the casualty insurance policy,
dependent upon the specific mitigation technology being considered
in the quote. Thus, for example, if the policy being quoted
includes coverage for premises/operations liability, the system
would display 222 a range of risk modifiers 223 for the selected
coverage. A risk modifier 223 is used to indicate where the
specific risk falls in relation to a base or average risk for a
given classification. In the present invention, the base risk has
two components, a first specific risk, historically associated with
a structure, and a second specific risk that mollifies the first
risk dependent on the technology utilized. The risk modifier 223 is
the result of the classification 217 of the various technologies
that are applicable to the building structure under consideration.
The loss control programs and technology that the insured
institutes, essentially influences the magnitude of the risk
modifier. In steps 224 the user selects one of the predetermined
risk modifiers for the selected coverage, and then documents 226
the reasons (e.g., loss mitigation technology or loss control
programs) for the specific risk modifier that the user selected. In
step 228, the process may be repeated for each type of coverage
dependent upon number or different technologies to be included as
part of the quotation.
[0050] In step 230, the system generates casualty insurance premium
amounts corresponding to a plurality of different insurance
attachment points. For each attachment point, the corresponding
premium amount generated by the system is based on, among other
things, a minimum premium amount associated with the SIC input in
step 204, the hazard rating code(s) selected in step 218, and the
two risk modifier code(s) selected in steps 223 and 224,
respectively. In a preferred embodiment, the premium amounts are
generated in step 230 from a table stored on the process 200
database 150.
[0051] Next, in step 232, the user selects one or more of the
attachment points generated in step 230 for quotation, and the
system then generates a quotation describing the policy being
quoted and stating a premium for the policy. The quotation is then
communicated to the insured.
[0052] After the quotation is issued in step 232, the system awaits
234 a response to the quotation. In the event that the quotation is
accepted, the system proceeds to step 238 where the policy is
thereafter bound. In the event a quotation 232 is not received in a
time specified in the quotation, the offer expires 236. The
quotation may also be accepted with proposed amendment 240, which
is there upon resubmitted via step 220 to re-quote the premium.
[0053] The method herein includes the automation of the collection
of technology strategies based in part on an automated decision
support tool, for objective evaluation of data relating to any
collection related to decisions or activities, and a historical
data warehouse, for comparison to all other non technology based
insurance underwriting.
[0054] The methodology herein includes systems which may perform
active polling and systems check utilizing communications systems
such as the Internet, telephone or broadcast transmissions in
combination with other forms of communications.
[0055] In FIG. 3a a classification device, such as neutral network,
is trained to learn an unknown function based on known inputs and
corresponding outputs. Once the neural network learns the unknown
function, it is able to generate outputs for other sets of inputs.
The invention classifies risk based upon comparing an insurable
interest, without the benefit of a technological enhancement,
against the same risk with a benefit of the technology have a risk
reducing value. In a preferred embodiment, a processor having one
or more inputs to receive data structures representing a first
unmitigated insurance underwriting risk pertaining to an identified
building structure and a second mollified insurance underwriting
risk on the assumption a certain technology will be employed. The
second mollified insurance underwriting risk data represents the
incorporation of a technology that aides in the reduction of the
financial consequences of property loss. In one embodiment, each
data structure typically forms a difference, referred to as a
weighted difference, between the first data field representing an
unmitigated risk, and the second data field representing a
mollified risk. The plurality of data structures generate weighted
differences to form a plurality of weighted outputs. At least one
of the weighted outputs represent a class for the mollified risk,
and the datum represents an input into a process that sums all the
weighted differences to generates a minimized risk for an insurable
interest under consideration. Such minimization processes are
well-known in the art of neural network design engineering.
Essentially, the processor calculates the weighted sum of one or
more inputs representing levels of risk, with and without the
technology to reduce the risk.
[0056] By way of example, referring to FIG. 3a having two (2)
inputs t.sub.0 303 and t.sub.1 304, and one output 305, represents
the simple output from two risk levels, one with and one without
the mitigating technology. The output is therefore:
W.sub.0*t.sub.0+W.sub.1*t.sub.1+K.sub.b>0 for a risk having two
different technologies, t.sub.10 and t.sub.2 incorporated into a
hypothetical insurable interest such as a building structure and
having a constant K.sub.b resulting from non technology related
losses.
[0057] In a preferred embodiment, FIG. 3a illustrates a neural
network classifier for categorizing and weighing risk data, for
identified technology used in a building. The classifier employs
modeling algorithms executed to generate data indicative of the
predictive risk.
[0058] The classifier categorizes and weighs risk data representing
for first unmitigated insurance risk 304 and a second mollified
insurance risk 303 based upon the use of certain technology. Inputs
303 and 304 are multiplied by a pre assigned weights, 308 and 309
respectively. At the summing junction 307, the data forms a
difference, referred to as a weighted difference, between the
unmitigated risk 304 and the mollified risk 303. A plurality of
such weighted differences are summed in a processor 302 having a
sigmoid transfer function such that when the output 305 is back
propagated at 306, the weighted difference generates a minimized
risk for a building structure under consideration. Essentially, a
processor operates to calculate the weighted sum of one or more
inputs representing levels of risk, with and without the technology
to reduce the risk. By way of example, consider the two (2) inputs
303 and 304 generating output 305 which data represents the
summation: W.sub.0*t.sub.0+W.sub.1*t.sub.1+K.sub.b>0 for a risk
having two different technologies, t.sub.1 and t.sub.2 incorporated
into a hypothetical building structure and having a constant K
resulting from a non technology related loss.
[0059] In order to adjust the premium base on mitigation
technology, such as required, for example, in step 223, FIG. 2, a
preferred method of classification as illustrated by the schematic
in FIG. 3b. The classification method comprises the steps of
generating a plurality of technology mitigation variables, wherein
each of the technology mitigation variables corresponds to an
underwriting classification commonly utilized to underwrite an
insurance policy. The classifier generates a plurality of actual
loss data from building structures that do not contain the
mitigation technology, wherein each of the actual loss data is
indicative of the actual loss of a plurality of policyholders. The
classifier further: (1) generates a plurality of premiums, wherein
each premium in the plurality of premiums is indicative of the
premium charged to one of a plurality of policyholders; (2)
generates a plurality of actual technology mitigation loss ratio
data, wherein each of the actual loss ratio data is indicative of
the actual loss ratio of one of the plurality of policyholders; (3)
generates depending upon the plurality of underwriting
classification plan variable values and upon said actual loss ratio
data, a plurality of predicted loss ratio data, wherein the
predicted loss ratio data, is indicative of a predicted loss ratio
of one of the plurality of policy holders; and (4) generates,
depending upon said plurality of underwriting classification value
data, said actual loss ratio data and said predicted loss ratio
data, which are indicative of a difference between the predicted
loss ratio and the actual loss ratio of one of the plurality of
policyholders.
[0060] FIG. 3b shows a neural network 300 used as an apparatus to
classify multiple technologies that support the mitigation of loss
consequent to the employment of technology. As well known in the
art, computational processing units, such as 310a through 310n and
330, are grouped into layers, wherein the output signals from
primary layers form input signals in secondary layers. The neural
network 300 inputs 312a and 312n, have a one-to-one correspondence
with technology mitigation data. Inputs 314a and 314n have a
one-to-one correspondence with underwriting data that does not
include technology mitigation data. These sets of data form the
input layer for processing unit 310a through processing unit 310n,
respectively. Processor 310a, receives input data 312a and 314a and
generates output data 316a representing a computation performed in
accordance with a transfer function, such as a summation function,
programmed or designed into the nonlinear element of the processor
318a-318n. A connection from a first processing unit 310a output
316sa is combined with a comparable one or more of a multiplicity
of processing units, such as by way of example 310n having an
output 316n, which forms input data to a third processing unit 330.
Optionally output 317 can be feedback or back propagated into the
processing units 310a-310n via switch 317.
[0061] In an alternate embodiment of the invention herein, FIG. 4
is a classifier 400 that assists in carrying out the foregoing
method, and assists in the performance of the step 213, 215, and
217, FIG. 2, to predict loss based on the mitigation of a
technology and thereby determine a requisite premium.
[0062] In the overview, a database of vectors 403 representing
combinations of technologies used in a construction of a building
are assigned magnitudes representative of their efficiency to
create a favorable loss experience as loss relates to one or more
insurable occurrences. Each vector in the database 403 is utilized
in a neural network as a training set. The training set calculates
a set of weights that when utilized with other actual proposed
constructions an a posteriori vector representing a potentially
insurable interest yields an approximate loss ratio.
[0063] The technology classifier contains a set of technology
mitigation vectors in database 402 as combination variables, each
specific to one building configuration. Essentially, the vectors
represent attributes of a plurality of technologies that reduce
potential property loss. The technology mitigation vectors each
contain a plurality of values. Note, that each of the technologies
extant in database 402, 403 may be stored on databases, such as 150
or 170, (FIG. 1). The vectors have an assigned premium or loss
premium coefficient related to a loss ratio mitigation factor.
[0064] The network 401 utilizes inputs 405, which have a one-to-one
correspondence with classification of mitigation technology values,
to form an input layer 410, which has as its input mitigation of
technology data stored in database 402 via line 415. Similarly, the
classifier has an output 425, which generates data indicative of a
predicted loss ratio. A single hidden layer 435 of neurons 440
couples input layer 410 to output layer 425.
[0065] Line 430b serves to adjust the weights in accordance with
the weight adjustment data resident vectors in database 403. The
controller 465 generates weight adjustment data on line 470, to
establish the optimum weights in the layers 410 and 435. Technology
mitigation data, which the controller 465 receives, are presented
as inputs to the network 401 via line 415 during what is commonly
referred to as the training mode for the neural network. The
controller 465 receives the output signals of the network 401 on
line 430b and compares them with the data indicative of the actual
loss ratio received on a line 460. The actual loss, weight
adjustment data is stored as a series of vectors, database 403 and
form the input 460 to controller 465. Controller 465 adjusts the
weights during the training process until the difference between
the output signal on line 430b and the desired output signal on
line 460 is reduced to a desired minimum.
[0066] When a input 450 received from database 451, representing a
technology configuration of a possible insurable, it is passed to
the controller 465 and then passed on to the neural network 401
whereby, the loss ratio is determined at 420, based upon the
weights as previously trained during the training mode.
[0067] In FIG. 5 a table 500 and FIG. 6 a decision tree classifier
600 are utilized to assign combinations of technologies, employed
in existing building structures to a classification, which then
permits an underwriter to establish a premium. In this manner data
structures representing the quantification of risk reduction
attendant a given technology or product can be chained into a
plurality of decision trees. In one aspect of the invention the
decision tree 600 includes a construction phase and a pruning
phase. The construction phase requires that the set of building
structures and corresponding combination of technologies be
recursively partitioned into two or more subpartitions, until a
stopping criterion is met and a classification assigned. The
decision applies a splitting criterion to every node of the tree.
These splitting criteria are determined by applying a predetermined
rule or function that an underwriter applies to eventually place
the applicant for insurance into a classification that is then
utilized as a factor in establishing the premium.
[0068] Each node may utilize any number of conventionally available
analytical techniques for producing a splitting criterion. For
example, each branch may be programmed to produce a weighted
average (W.sub.1-N) that may subsequently be applied in successive
nodes and branches to influence the final risk classification. On
the other hand, the decision criterion might be a simple binary
decision, where an attribute either exists or does not exist in the
data set.
[0069] In one aspect of the invention, a processor (unshown) is
programmed to impose a decision criterion based upon whether a set
of technologies exist in a particular type of building structure.
As such the invention is also a method for classifying risk
reduction technology comprising the steps of: storing data
representing a training set in a memory, the data including a
multiplicity of entries each having a plurality of attributes, such
as type of construction, combinations of technologies and other
relevant factors; and classifying the training set by building a
decision tree based on the attributes of the training set entries.
Thereafter, a method employed in classifying an underwriting risk
utilizes the decision tree based on the attributes of the training
set. FIG. 5, table 500 tabulates hypothetical table entries 510 of
building structures, each of which are representative of a
structure type 520. A structure type 520 may, by way of example, be
a wood frame construction and in another instance be a masonry
construction. Each of the differing type structures may include
combination of technology 530 indicative of risk reducing
technologies. When the type structure 520 is utilized in
combination with a combination of technology 530, a class 540 and
optionally a subclass 550 are assigned. Subsequently, actual type
structures having the identical combination of technology will be
classified in accordance with the representative table 500 as
falling within one of the class 540 and subclass 550. The foregoing
table 500 may be utilized as data for any number of expert system
analysis. For example, in an alternate embodiment the data
contained in table 500 would be utilized as input to a multivariate
statistical analysis or to create a continuous function that would
lead to values that selectively result in specific underwriting
choices, exclusions or excess premium charges.
[0070] In FIG. 6, the partitioning begins at node 605 where the
combination of technology 530 containing technologies 1, 2, 5, 3, 7
abbreviated TC (1, 2, 5, 3, 7) from Table 500, leads to a
bifurcation. A "yes" indicating that such a combination exists in
the table entry 510, leads to node 610. A "no", indicating that
such a combination does not exists in the table entry 510, leads to
node 650. At node 610, the table entry 510 that meet the criteria
established at node 620 are listed and they are subjected to a
second set of criteria, as for example, whether the type structure
520 falls below an ordering criteria 612 established by yet another
criteria, which for purposes of this description may be, a function
of organizing type structures 520 on the basis of imperviousness to
hurricanes, fire or other catastrophic events. In the present
example, structures falling below the letter designation "H" will
lead to node 620. At node 620 ordering criteria 613 classifies
TE(1,2) into two classes at node 624 and node 630, respectively. A
classification of TE(2) is established at node 615 designated as
Class A1A. In a like manner, a "no" decision at node 620, leads to
node 624 and a classification of A1B for TE(1). At node 650,
TC(2,3,4,0,9) is subjected to whether the combination of
technologies include TC(2,3,4,0,9). If they do, then those table
entries are culled out for examination at node 660. At node 660 the
table entry TE(3,5,6,7) are tested against whether the meet the
criteria 614, falling before "D" in the type structure 520 column.
If the criteria 614 is met, then a further node 670 tests each
table entry 510 against a criteria 616 relating to the type
structure being less than "B" in the table 500. Other branches
designated by node 560 may produce corresponding direction changes
based upon a test TS (L, D) 214 leads to node 672. A test TS (L, C)
leads correspondingly to either the node 680 or a classification
Class A, E 665. Finally at node 635 a class A2B is assigned to
table entry TE(5). Similarly at node 690 a type construction
TC(1,2,4,3,0) is simply tested against the existence of a table
entry 510. If a table entry 510 does not exist then the node 665
assigns a class A5A to the table entry or entries as may be the
case. Various null paths such as null 680, may be further assigned
underwriting significance as the particular case may require. The
various paths that the decision tree follows in any particular
instance represents a pruning of those "risk branches" that were
amenable to insurance coverage due to risk reduction mechanisms
being employed.
[0071] As will be apparent to those skilled in the art of computer
software, the programming language, and where and on what computer
the software is executed is a design choice. The foregoing
description of decision tree 600 as configured is by way of
illustration and example only and is not to be taken by way of
limitation, the spirit and scope of the present invention being
limited only by the terms of the appended claims.
[0072] A method of underwriting insurance by taking into account
technologies that militate against loss may comprise the steps of:
identifying a technology that mitigates a risk associated with a
property loss for which an insured purchases insurance; and
providing an insurance policy that accounts for the diminution of
risk. In an embodiment, the method may further include the step of
advising the prospective insured to obtain the technology. In an
embodiment, the method may also include the further step of
advising the prospective insured about the cost benefits of
obtaining the technology. In an embodiment, the method may also
include the further step of providing a specification of best
practices to mitigate losses through the application of a
technology.
[0073] A method for underwriting insurance by taking into account
technologies that militate against loss may comprise the steps of:
maintaining a database identifying a plurality of technologies that
reduce risk of loss to an associated building structure;
identifying a building structure comparable to the associated
building structure, that requires insurance; calculating the risk
of loss related to the building structure; and accounting for the
risk reduction resulting from incorporation of at least one
technology into the building structure; and creating an insurance
policy for the structure based upon the incorporated risk reduction
technology. In an embodiment the method may also include the
further step of polling the insured interest to determine its
compliance with incorporation of at least one technology into the
building structure.
[0074] A system for underwriting insurance by taking into account
technologies that militate against loss may comprise: a means for
classifying risk input data, the means include a processor having
one or more inputs to receive one or more data structures
representing a first plurality of unmitigated underwriting risks
pertaining to an identified building structure and a second
plurality of mitigated underwriting risks based on the assumption
one or more specified technologies will be employed in the building
structure, a logical association between the first and second data
structure representing the incorporation of a technology that aides
in the reduction of a casualty property loss for the building under
consideration for insurance wherein, each data structure combined
forms a weighted difference between the first unmitigated risk and
the second mitigated risk, and wherein a plurality of weighted
differences represent an underwriting class for the unmitigated
risk/mitigated risk combination to generate a minimized risk for a
building structure under consideration.
[0075] A system for underwriting insurance by taking into account
technologies that mitigate against loss may comprise: one or more
decision trees, each branch of which produces a weight representing
an underwriting risk specific to an insurance coverage and premium.
In an embodiment of the system, a continuous function may be
utilized to assign the weight to produce risk values representing
underwriting choices, exclusions, and premium charges to an
insurance policy for combinations of property and risk reduction
technology. In an embodiment, the output means may be a device to
print a specification of best practices for loss mitigation based
on the subject building.
[0076] A computer system for evaluating the insurability of a
potentially insurable risk, may comprise: a means providing first
and second data bases; a means for storing information relating to
the potentially insurable risk in the first data base; a means for
storing information relating to the potentially insurable risk
mitigated by a technology in the second data base; a means for
evaluating the information stored in the first data and second data
base and for identifying additional elements of information
required for evaluating the potentially insurable risk, including:
means for assigning a weight to at least one of the selected
elements of information from the first data base on the basis of a
relationship between the elements of information in the first data
base and corresponding elements of information in the second data
base; a means for associating the weights to calculate at least one
risk classification for the potentially insurable risk from the
weights assigned to the elements of information from the first data
base; and a means for displaying at least one at least one risk
classification. In an embodiment, the means for assigning a weight
includes one or more expert systems.
[0077] A computer system for managing insurance based upon
technology utilized in a structure may comprise: a plurality of
workstation processor means for managing one or more insurance
accounts; a database processor means for storing, at least one
element of data which depicts technology that mitigates insurance
risk, the database processor means being interconnected and
responsive to each of the plurality of workstation processor means;
a file processor means for managing at least one element of data
which represents technology that mitigates insurance risk, the file
processor means being interconnected and responsive to each of the
plurality of workstation processor means; an output means for
producing documents in at least one of text, graphics, and
electronic transfer mode, the output means being interconnected and
responsive to each of the plurality of the workstation processor
means; and, an input means for receiving at least one input, of
which represents technology that mitigates insurance risk, into the
computer system, the input means being interconnected and
responsive to each of the plurality of workstation processor means;
and, a software means for configuring each of the plurality of
workstation processor means, database processor means, file
processor means, output means, and input means. In an embodiment of
the system, the output means may be a device to print an insurance
policy based on an insurable interest.
[0078] A method of underwriting insurance may comprise the steps
of: maintaining a data base identifying risk mitigation technology
and corresponding loss mitigation values; and scanning periodically
an insurable interest to identify sensors and corresponding sensor
data associated with risk mitigation technology; and comparing
sensor data to the risk mitigation technology and corresponding
loss mitigation values; and controlling a printing device to print
an insurance policy.
[0079] A system for identifying matches between risk mitigation
technology installed at an insurable interest and a risk mitigation
technology listed in an insurance policy may comprise: a database
storage means; a processor programmed to: maintain the database
storage means wherein is identified installed risk mitigation
technology; audit the insurable interest to identify sensors and
corresponding sensor data associated with installed risk mitigation
technology; and compare corresponding data to identified installed
risk mitigation technology stored in the database storage means;
and control a printing device to print a report on the operational
status of the identified sensors.
[0080] A computer readable medium may have stored thereon one or
more data structure selected from the group comprising of: a first
field containing data representing an indication of the conformity
to a prescribed level of risk due to the use of risk mitigation
technology; and a second field containing data representing the
address of an insurable interest associated with the first field
data; and at least one field containing data representing a premium
adjustment related to the utilization of the risk mitigation
technology.
[0081] A neural network may comprise: an input layer, an output
layer, and at least one hidden layer, operable to produce outputs
from the output layer when inputs are supplied to the input layer
of the neural network, the neural network having been previously
trained in accordance with training exemplars of loss mitigation
due to the incorporation of certain technologies.
[0082] A method for performing risk analysis utilizing a neural
network may comprise the steps of: collecting data on a first
combination of technologies; and applying the data to an input
layer of the network; and training the network by optimizing
weighted connections of the network to an underwriting criteria
that decreases the risk of an insurable interest such that weights
are determined; and applying a second combination of technologies
to an input layer of the network; and classifying the second
combination of technologies into ordinal values and categorical
values, such that the classification represents a premium
estimation.
[0083] A method for classifying risk reduction technology may
comprise the steps of: storing data representing a training set in
a memory, the data including a multiplicity of entries each having
a plurality of attributes, such as a type of construction and a
combinations of technologies; and building a decision tree based on
the attributes of the training set entries thereby classifying the
training set by. In an embodiment, the method may include a further
step that includes classifying an underwriting risk using the
decision tree based on the attributes of the training set.
[0084] Although the present invention has been described and
illustrated in detail, it is clearly understood that the same is by
way of illustration and example only and is not to be taken by way
of limitation, the spirit and scope of the present invention being
limited only by the terms of the appended claims.
* * * * *